negotiating peace in kosovo

This article was downloaded by: [The University of Manchester Library]On: 17 December 2014, At: 08:29Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK

International Interactions: Empirical and TheoreticalResearch in International RelationsPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/gini20

Negotiating peace in KosovoBrian Efird a , Ambassador Peter Galbraith b , Jacek Kugler c & Mark Abdollahian da Decision Insights, Incorporated and Claremont Graduate University , 60 East 42nd Street,Suite 514, New York, NY, 10165b The National War Collegec Claremont Graduate University, School for Politics and Economics , 170 East 10th Street,Claremont, CA, 91711–6163d Decision Insights, Incorporated , 60 East 42nd Street, Suite 514, New York, NY, 10165Published online: 09 Jan 2008.

To cite this article: Brian Efird , Ambassador Peter Galbraith , Jacek Kugler & Mark Abdollahian (2000) Negotiating peacein Kosovo, International Interactions: Empirical and Theoretical Research in International Relations, 26:2, 153-178, DOI:10.1080/03050620008434964

To link to this article: http://dx.doi.org/10.1080/03050620008434964

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in thepublications on our platform. However, Taylor & Francis, our agents, and our licensors make no representationsor warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Anyopinions and views expressed in this publication are the opinions and views of the authors, and are not theviews of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should beindependently verified with primary sources of information. Taylor and Francis shall not be liable for any losses,actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoevercaused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/loi/gini20

http://www.tandfonline.com/action/showCitFormats?doi=10.1080/03050620008434964

http://dx.doi.org/10.1080/03050620008434964

http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/page/terms-and-conditions

International Interactions © 2000 OPA (Overseas Publishers Association) N.V.VoL 26, No. 2, pp. 153-178 Published by license underReprints available directly from the publisher the Harwood Academic Publishers imprint.Photocopying permitted by license only part of The Gordon and Breach Publishing Croup.

Printed in Malaysia.

NEGOTIATING PEACE IN KOSOVO

BRIAN EFIRD

Decision Insights, Incorporated andClaremont Graduate University

AMBASSADOR PETER GALBRAITH

The National War College

JACEK KUGLER

Claremont Graduate University

MARK ABDOLLAHIAN

Decision Insights, Incorporated

(In final form March 10, 2000)

We systematically analyze and forecast key developments before and during the KosovoCrisis. Prior to the air campaign, we anticipated the breakdown of negotiations at Rambouilletand the subsequent initiation of hostilities. Without Russia as a mediator, we predicted thatMilosevic would engage in military activity and strengthen his domestic control. As the warwas being waged, we took a second look at settlement opportunities. Russia was identified asthe key mediator needed to reach a settlement We show that a European-led settlementoffered promising prospects for peace. Finally, we demonstrate that the settlement achievedafter the air war was not much different than the settlement that could have been achieved atRambouillet The settlement on Kosovo offers only short-term stability. To gain long-term

153

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14

154 B. EFIRD ETAL.

stability, the Serbians must either agree to the existing ethnic makeup in Kosovo or allow it tobe partitioned. Otherwise, conflict in this region will likely revive.

INTRODUCTION

The study of international politics has changed dramatically in the lastthree decades. The fundamental contribution of more formalized andempirical approaches to the field is the reduction in the number of theor-etical structures that can be used to account for a single phenomenon.Improvements in the way we empirically measure and test new proposi-tions can lead to a more systematic and scientific understanding of worldpolitics. However, the substantive impact of this expanding knowledgeremains in the classroom. Very few new insights are used to affect policy.The long shadow of misapplications of political science to policymakingduring the Vietnam War demonstrates that it is both dangerous andimproper to allow incomplete results to guide policy, only to discover thatthe vast simplification required for a scientific assessment obscuresimportant elements in the real world. However, academic inaction is notthe solution. When analyses that can reduce the likelihood of war are notexplored, when opportunities to strengthen negotiations are not exercised,when options that could prevent the escalation of crises are not takenadvantage of, then silence is as dangerous as advocacy.

Here we explore the process of negotiations between the US, Yugosla-via, and the rest of the international community using the systematicexpected utility approach pioneered by Bruce Bueno de Mesquita (Buenode Mesquita and Stokman, 1994; Bueno de Mesquita, Newman andRabushka, 1996; Bueno de Mesquita, 1997). Work derived from this modelsuggests that the outcome of negotiations can be anticipated and in manycases improved (Kugler and Feng, 1997).1 To test this notion, we appliedthe expected utility approach to the Kosovo crisis as it was evolving. Weidentified the autonomy of Kosovo and the deployment of troops to overseea settlement on autonomy as the two key issues within a very complicatedset of negotiations. Our objective was to ascertain if it is possible to accountfor the evolution, escalation, and termination of conflicts.

In the first part of this paper we present analyses that accuratelyanticipated the breakdown of negotiations at Rambouillet and foresaw thesubsequent initiation of hostilities. In the second part, conducted at theheight of the air war, we anticipated the terms of settlement that led to anend of NATO bombing in Kosovo. The analysis identified Russia as a keymediator. Finally, in systematic comparisons we demonstrate that the air

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14

NEGOTIATING PEACE IN KOSOVO 155

war achieved an outcome not much different from the one that could havebeen gained at Rambouillet without the bombing campaign.

This paper presents time sensitive forecasts and contrasts them to actualoutcomes. Using the expected utility decision making approach, we ana-lyze the bargaining process at two points in time. Our first cut focused onevents during the Rambouillet negotiations. Our second cut extended thisanalysis to the outcome of the air war.

As events unfolded, the first cut was based on data collected fromAmbassador Peter Galbraith on February 10, following the initiation oftalks in Rambouillet, France on February 6,1999. The results of this ana-lysis were presented on February 17 at the National War College, priorto the breakdown of talks at Rambouillet on February 26.2 AmbassadorGalbraith also provided data for the second cut on April 25, and the analysiswas completed on April 29. All analysis was completed before the agree-ment to cease bombing was implemented on June 10. Before discussing theparticulars of this analysis, let us review the approach taken in this paper.

A DECISION MAKING APPROACH TO NEGOTIATIONS

The decision malring model we use to evaluate the Kosovo negotiations,often called the expected utility model,3 is based on the work of Bueno deMesquita and his colleagues. The mathematical formulations for thismodel have been described in detail in a recent special issue of Interna-tional Interactions edited by Kugler and Feng (1997). We thus confineourselves to a summary of the logic and provide only a brief explanationof the relevant results it produces.

The expected utility approach assumes that all stakeholders in nego-tiations are rational, net gain maximizers that adjust their actions inresponse to perceived gains and losses. Such stakeholders do not behave ina uniform manner, rather under similar objective conditions they act ontheir own perceptions of reality. The model assumes complete informa-tion, but the perceptions of outcomes are affected by the risk propensity ofcompeting actors. The model is based on two well-established micro-economic and political theorems. First, it relies on insights from Black's(1958) median voter theorem where stakeholders in any bargainingsituation are treated as weighted voters.4 This theorem shows that themedian position is the winning outcome under majority rule.

The expected utility approach also utilizes Bank's theorem on themonotonicity of expectations to anticipate the intensity of conflict. Gen-erally speaking, conflict escalates in political disputes when the "votes" of

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14

156 B. EFIRD ETAL.

viable stakeholders are far apart on the same issue. Proximity generatessupport, leading to effective negotiations. When most stakeholders are farapart or they split into coherent groups, then a conflict is likely and anyagreement will be tenuous. Under such conditions, negotiations usuallyfail or an agreement, if reached, is violated.

The model requires only five pieces of information for its computa-tions. Ambassador Peter Galbraith, who is intimately familiar with thedispute over Kosovo, provided the required data:5

The central issues of disputeThe stakeholders trying to influence the resolution of the issuesThe range of policy preferences on the central disputed issues and thestated position of each stakeholderThe relative influence of each stakeholderThe salience of each issue to every stakeholder.

Stakeholders are the domestic and foreign actors who have the capabilityto influence the outcome of the Kosovo crisis. Note that in the Appendix,the set of actors analyzed for these issues include individuals, organiza-tions, and countries. The expected utility model applies to bargainingamong individuals, countries, or individuals and countries. As long as eachstakeholder asserts a unified opinion in negotiations, groups at differentlevels of aggregation can be considered at the same time.

Policy preferences identify the stated positions of stakeholders on thedisputed issues at the time of the data collection. Unlike economics, thecurrency of politics changes from case to case. For this reason, an effectiveanalysis requires expert knowledge of the issues that are determinant ineach crisis. Effective forecasts are only possible if the expert knows thecritical issues of debate and can place them on continuous scales. Feder(1995) provides a systematic assessment of the performance of expert-generated data with the expected utility model using a large number ofcases. He finds that while experts may disagree with the predictions of themodel, their data tend to vary only slightly and do not produce appreciabledifferences in the model forecasts (Feder, 199S). This congruence suggeststhat the results of the model are robust.

Relative influence reflects the potential resources that stakeholders canbring to bear on each issue, relative to the resources of other stakeholders.Finally, salience indicates the willingness of stakeholders to focus theirenergy and resources to achieve their preferred outcome on the issue.Together, potential resources multiplied by salience represent the actualpower of stakeholders during negotiations over each issue.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14


Using this information exclusively, the expected utility model providesdetailed information about the stakeholders and the issue that was usedto anticipate the outcome of negotiations during the Kosovo crisis. Themodel produces five important pieces of information that can be used toanticipate the outcome of political interactions, negotiations, and conflicts:political interactions between every pairwise combination of stakeholders,willingness of each stakeholder to compromise or escalate the dispute,length of negotiations, changes in stakeholders' policy positions over time,and realized and unrealized negotiation offers and tradeoffs for each stake-holder across different issues. '

First, the model provides a means to estimate the outcome of politicalinteractions. Under majority rule, the model uses the median position,weighted by the resources available to each stakeholder and modified bythe salience attached to the issue, as the likely outcome. While the medianposition is useful in elections, majority rule does not apply when "veto"stakeholders, identified by the expert, can unilaterally prevent the imple-mentation of an agreement negotiated by the majority.6 When veto playersare present, as was the case in the Kosovo crisis, negotiated agreementsrequire that veto players agree on the same policy position.

Second, to anticipate the willingness to compromise or escalate a dis-pute, the model distinguishes between those outcomes that will be con-tested and those that will be accepted by competing parties. Whenstakeholders agree on a position, the intensity of conflict is reduced and asettlement is stable. However, if stakeholders are forced to adopt a positionthey disagree with, conflict is anticipated even after signing. Thus, themodel permits discrimination between cooperative and conflictual nego-tiations as well as compliance after signing an accord.

Third, negotiations proceed in rounds, which reflect the series of meet-ings or discussions in which stakeholders seek an outcome. Over time,stakeholders exert their influence upon each other, and successive roundsreflect the anticipated changes in stakeholder positions resulting from suchinteractions. Rounds may occur frequently, as is the case during last minutenegotiations, or infrequently, as is the case when the parties have littledirect contact This allows one to anticipate the length of negotiations orpolicy disputes.

For each round the model calculates how stakeholders adjust in reactionto demands from other stakeholders. The figures presented in this paperreflect the anticipated changes in stakeholder positions during the crisis,based on the calculations described in the literature referred to above.Specifically, the model identifies which stakeholders shift position in eachround, and which stakeholder has influenced them to adopt a new position.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14

158 B. EFIRD ETAL.

To the degree that such evaluations reflect reality, the model anticipatesthe political interactions, shifts, and options during negotiations.

Fourth, the model identifies situations where the stakeholders fail torecognize that they can advance their own position by making a credibleoffer to a third party. Depending on the political landscape of negotiations,different stakeholders will be able to credibly apply pressure to each other.The model helps identify those players who can successfully push througha proposal, the degree of influence they have, and with whom they havesuch influence. Using this information, the model identifies a credible,acceptable "unseen opportunity" that the stakeholders themselves/a// torecognize. Such insights can be used to construct negotiation strategiesthat alter the political landscape and potentially resolve the conflict.

In addition to the analysis provided by the expected utility approach ina single dimension, the predicted positions of stakeholders can be com-pared across two issues. This notion is formalized using the solutionconcept of winsets, developed by Weingast (1979) and elaborated byHinich and Munger (1997). Abdollahian and Alsharabati (1995) detail theconnection between expected utility calculations and winsets. Winsetsdefine the two-dimensional space where credible tradeoffs between issuesmay occur, identifying where a stakeholder's position on one issue can betraded for policy gains on another issue. A potential agreement is nolonger a single position with varying degrees of support, but rather theoutcome is now an area or solution space that represents a range of cred-ible tradeoffs. Since the winset solution space represents possible agree-ments that are acceptable to all parties, any solution is of equal value. Thusall proposals within the winset space are equally credible. Let us proceedto review the results of our analysis.

PART I: THE RAMBOUILLET TALKS

The war in Kosovo was the culmination of a long-standing dispute. InOctober 1998, US envoy Richard Holbrooke led negotiations that producedan agreement between the United States, NATO, and Yugoslavia thatallowed for NATO aerial surveillance and unarmed Organization ofSecurity and Cooperation in Europe (OSCE) troops to monitor human rightsabuses. Violations of this agreement led to the subsequent negotiations atRambouillet As these negotiations unraveled, the Serbian forces in Kosovoinitiated and then escalated the intensity of their ethnic cleansing.7 Theunderlying cause of this dispute was the failure of the Yugoslavian politicalsystem to effectively accommodate ethnic divisions within the region.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14


We gathered the data for this analysis on February 10, 1999, whileSerbian and Kosovar representatives were meeting with European and USnegotiators in Rambouillet (for details, see the Appendix). AmbassadorGalbraith identified the autonomy of Kosovo and the deployment of troopsto oversee a settlement on autonomy as the two key elements within a verycomplicated set of negotiations. He argued that settling on the degree ofautonomy would remove the major impediment for cooperation betweenSerbians and Albanians. Furthermore, a deployment of troops could ensurecompliance with the agreement on autonomy, and would prevent the useof violence by both sides.

The model predicted the changes in positions of all stakeholders on eachof these issues. The results are presented in Figures 1 and 2. The verticalaxis in these figures reflects the position each stakeholder takes on the issue(the definitions and numerical ranks are in the Appendix). Note that thescale focuses on the substantive policy alternatives identified by Ambas-sador Galbraith. The horizontal axis reflects the stakeholders who influencethe issue. The final axis reflects the shifts in the positions of stakeholders ineach round of negotiations. The median position at the far right identifiesthe majority position as a reference. The veto players that will determine

Fullp. Independence• ' H for Kosovo

Autonomy

Limited"¿if l ¡ Autonomy

Round 1 -

Round^ . • J Í - . - S ' * NO Self-Rule

II

Figure 1 Rambouillet Forecast: Stakeholder Position Shifts on Autonomy ofKosovo

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14

160 B. EFIRD ET AL.

the outcome in this analysis are the United States' Government, the KosovoLiberation Army (KLA), the Democratic League of Kosovo (or LDK,represented by its leader, Ibrahim Rugova), and Milosevic.

Figure 1 focuses on the degree of Kosovar autonomy. The scale ofpositions that stakeholders may adopt range at uneven intervals fromindependence, full autonomy, limited autonomy, to no self-rule. The initialpositions were defined by Ambassador Galbraith, subsequent policy posi-tions in round 2 and on are derived by the expected utility decision model.

Note mat after two rounds of negotiations the model anticipates that theprimary representatives of Kosovo (the LDK, the KLA, and other Kosovarstakeholders) will give up demands for full independence and accede tothe US position on autonomy. This is exactly what happened during lastminute negotiations following Rambouillet in Paris on March IS.

Figure 1 also shows that Milosevic and the Yugoslav Army presented arelatively moderate front by Serbian standards. However, a review ofsubsequent rounds shows that Western pressures for further concessionswould be resisted. As the record now shows, Serbians even rejected themore moderate demands by Russia, Macedonia, and Montenegro forlimited autonomy. We therefore anticipated, several weeks before the finalbreakdown of negotiations, that the Serbians would reject a negotiated

. ï , ~ r r , .NATOi , | , I f Troops

VJ_ No Foreign_ Presence

Figure 2 Rambouillet Forecast: Stakeholder Position Shifts on Troop Deployment

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14


agreement that had already been approved by the Kosovars. This assess-ment proved accurate.

A similar story emerges from the analysis of troop deployment—anissue of intense sensitivity to Serbians. Figure 2 shows that at one extreme,stakeholders advocated a force composed of NATO peacekeepers thatexcluded the Serbian military. At the other end, no foreign presence wasallowed by Yugoslavia and Serbian troops remained. The status quoinvolved unarmed OSCE peacekeepers in compliance with the Holbrooke-Milosevic compromise. Intermediate points represented a mixture ofNATO, Serbian, and other troops.

The forecast regarding the bargaining over troop deployment showsthat all Serbian factions rejected armed peacekeepers to guarantee auto-nomy in Kosovo. Western stakeholders initially demanded NATO troopsonly, but the model anticipated that these demands would be lowered.However, the model also anticipated that any compromise proposed by theWest would fall far short of the rigid demands made by Serbians. Thus,despite early optimism, we anticipated that the Rambouillet negotiationsand the follow up in Paris would fail before any agreement was signed.8

As noted in Figure 2, most stakeholders take strikingly different positionson this issue. Had a resolution been agreed on at Rambouillet, we wouldobserve a very close alignment of stakeholder positions.

MISSED OPPORTUNITY: A POSSIBLE RUSSIAN SOLUTION

In addition to anticipating the Rambouillet outcome, we also assess solu-tions the model indicated were viable and could have been made, but werenot pursued.9 The objective was to identify whether an outcome, short ofwar, could have emerged from negotiations. Our analysis shows that at theRambouillet talks, the Russians had an opportunity to propose a viablesettlement that required a tradeoff linking troop deployment with Kosovarautonomy.10

To identify this missed opportunity, the analysis produced by themodel suggested that the Russians would be able to pressure the Serbianstakeholders in round 2 of negotiations. Implementation of this complexnegotiated settlement would have required two steps:

1. The Russians would have to lead the dialogue.2. The US government and France would have to shift positions and

strongly support the Russian proposal for limited autonomy (US andFrench salience increases to 100).

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14

162 B. EFIRD ETAL.

Had steps (1) and (2) been followed at Rambouillet, the model indicatesthat in the second round, the Russian delegation could have made a jointproposal acceptable to the KLA, the LDK, and Milosevic. In such nego-tiations, French and US government approval would have been necessarybut their direct involvement would have been an impediment to resolvingthe crisis. A Russian-led "solution" would have required the United Statesand France to stand by and allow the emergence of an agreement halfwaybetween full and limited autonomy (there is no winset if the US andFrance are included).11 Figure 3 identifies a winset on these two issuesbetween the critical actors in the analysis. This indicates that the criticalparties would have concurrently agreed to a troop guarantor that includeda mixture of NATO, Russian, and OSCE monitors if they could tradeoffwith gains on the autonomy issue, had such an offer been proposed.

In Figure 3, the leader of the Serbian faction (Milosevic) and theKosovar faction (the LDK and KLA) agree to a viable compromise.12

Once the key stakeholders reach this agreement, they do not defect Insum, this complex, sensitive negotiation represents the "best" option thatcould have been achieved in Paris. Note that the US ideal point withinthe winset is at the opposite edge of the solution space than that of the

N A T O -Troops

No ForeignPresence

NoSelf-Rule Autonomy

Autonomy of KosovoA Groupt H Status Quo

Autonomy * FullIndependence

Figure 3 Rambouillet Simulation: Russian-led Solution in Two Dimensions

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14


Russians and Milosevic. However, the presence of a winset suggests that theUS was willing to accept a substantial compromise on Kosovar autonomy togenerate an agreement Milosevic, on the other hand, does not have to shiftfar from his ideal point within the winset to the Russian proposal.

Recall that all options within a winset are of equal value to the stake-holders. Yet, note that we selected the optimal point within the winsetfrom the Russian perspective since this is the group that proposes the deal.The United States would have favored a different point in the winset, butthey were in no position to propose it as a US concession would have beenrejected. Note that Milosevic would have made a proposal almost identicalto the one made by Russia and subsequently would have likely supportedthis deal.

Even though such a complex solution would have left the West lessthan totally pleased, a Russian solution could have emerged from Ram-bouillet. The Serbians would have granted more autonomy to Kosovarsthan they were willing to accept without a tradeoff, and would haveallowed limited armed troops in the Kosovo region that they previouslyhad adamantly rejected. The West would have gained a modicum ofstability, but its troops would not have been able to ensure the full imple-mentation of this deal. Russia would have become the critical stakeholderwhose actions decided whether autonomy was limited or extended in thelong term. Whether this potential Russian solution would be palatablefrom a strategic US perspective is another question. Regardless, our ana-lysis indicates that the Rambouillet negotiations could have produced apotentially stable agreement. Policy makers either did not identify thisopportunity or chose to bypass this option for their own reasons.

PART II: THE AIR CAMPAIGN

An important and unresolved policy question is whether the military actionand its aftermath resolved the underlying tensions and conflict withinKosovo.13 Following the collapse of the Rambouillet negotiations, onApril 29 during the initial stages of the NATO air war, we gathered asecond wave of data (see the Appendix for details).14 The issue now facingthe stakeholders was the degree of autonomy for Kosovo. AmbassadorGalbraith no longer felt that troop deployment could be independentlyassessed. Because of the war, an increase in the degree of autonomy nowwould have implied increasing numbers of foreign troops needed toguarantee it. The range of positions on this scale is the same as above.Figure 4 shows the revised set of stakeholder positions and shifts.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14

164 B. EFIRD ETAL.

limited Autonomy

Very limited Autonomy

No Self-Rule

Figure 4 Wartime Forecast: Stakeholder Position Shifts

During the war, each coalition is closely aligned.15 Milosevic and Ser-bian stakeholders, except for the Bosnian Serbs, take an uncompromisingand coherent stance that rejects autonomy for Kosovo. Given this align-ment in the Serbian position, Milosevic is not expected to face internaldissent within Yugoslavia. Short of outright defeat, the model suggests thatYugoslavia would forcefully expel the ethnic Albanian population fromKosovo and would reject discussions on Kosovar autonomy. To ensure thisoutcome, the Serbian side was willing to endure the air war.

Western stakeholders, on the other hand, were initially less coherent intheir policy stance but then coalesce to a position of limited autonomy. InRound 1 these stakeholders demand full autonomy and no displacement ofethnic minorities. However, after the air war starts (by Round 2), theNATO coalition is willing to accept limited autonomy in order to achievepeace. Note that despite their preponderance, Western forces lower theirdemands, seeking a negotiated solution to the crisis. However, initial con-cessions by the NATO members are insufficient to placate the Serbians.Given this impasse, we anticipated that the air war was likely to continue!Thus, without additional interventions, Western flexibility would notlead toa resolution of this crisis. The unfolding of events confirmed ourevaluations. The next section illustrates the strategies the model identifiedthat could produce a stable outcome.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14


THE PROTRACTED CONFLICT OPTION

The rationale for bombing Yugoslavia rested upon imposing a sufficientenough sanction that would directly deter Milosevic's actions in Kosovo.Indeed, NATO planners believed that there was a significant chance thatMilosevic would withdrawal from Kosovo within the first three days of thebombing campaign (Priest, 1999d). We were able to assess the validity ofthis belief with the expected utility model. To simulate the effects of a pro-tracted air war, we first steadily reduced resources available to Milosevicand other Serbian stakeholders to establish the point at which sufficientcosts from die aerial campaign would have been imposed to produce asettlement. We found that simply imposing losses on the Serbian forceswould not have fostered concessions on the part of Milosevic.16 In short,our analysis showed that the air war alone was unlikely to end this conflict

Given this result, we concentrate on a negotiated solution that wouldbring the disputants together beyond the air war. Once more we identifieddie Russians as the critical stakeholder that could have brokered a stabledeal.17

The Russian Option

Even during the siege by NATO air power, a Russian-led option againemerged as a viable alternative to reach a negotiated solution in Kosovo.Recall that this alternative was identified during the last stages of theRambouillet peace talks. As we were conducting our analysis at the endof April, Russian envoy Viktor Chernomyrdin was attempting to forge acompromise between Serbian and Western demands. Our analysis indic-ated that, to succeed, a Russian-led initiative should have evolved slowly;it also required strategic collaboration with the United States. With themodel we determined that the following steps could have generated astable negotiated outcome:

1. The anti-war coalition in the US Congress needed to moderatelyincrease the intensity of its opposition to the air war (salienceincreases to 60).

2. In the third round, the US had to join and approve the Russianproposal.

The delays indicated in step (2) were needed because in the early stages ofthis negotiation, the Russian envoy did not have sufficient flexibility to

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14

166 B. EFIRD ET AL.

Limitad Aaleaomy •

Voy Uniud Asloaaay

NoSdf-R.l.

R e d 7

Figure 5 Wartime Simulation: Key Stakeholder Position Shifts with the RussianProposal

persuade the Serbian side to acquiesce to the compromise proposed.Continued US pressure was needed to make the Russian alternative lookincreasingly palatable to the Serbian contingent We accurately concludedthat negotiations would be lengthy and complex.18

Figure S displays the position shifts for the veto players only, whichwere of course needed to reach a negotiated settlement. The model sug-gests that the anti-war coalition in the US Congress had to walk a fine linebetween opposition and advocacy. If it protested the air war too strongly, itwould undermine the credibility of the NATO military threat. On the otherhand, if it protested US military activity too weakly, then Serbian stake-holders would be unwilling to negotiate a compromise. However, after thecrisis, we found no evidence that Congress played the role we anticipated insettling the air war. Such carefully calculated protestations may or may nothave taken place.

The model also identified that this would be a very slow-evolving andconfiictual settlement. We surmised that the air war would continue duringnegotiations. Again, the Russian initiative required delicate negotiations tosucceed. Milosevic was initially reluctant to accept this compromise andwas expected to accept it only after protracted negotiations (round 5). Atthat point however, he was expected to underestimate his own power andaccept higher levels of autonomy for Kosovo than was necessary. Theoutcome is stable since veto players as well as a majority of stakeholders

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14


adopt the same position. Such agreement ensures that the deal, oncereached, will be implemented effectively. This stable result foreshadowedthe unexpectedly orderly withdrawal of Serbian troops after the agreementwas reached.19

The results of the air war are consistent with our expectations. Asindicated by the specific stakeholder position shifts in Figures 4 and S, weaccurately anticipated that the KLA and LDK. would cave into the demandsof the West Note the we correctly foresaw mat the Russians would haveto be a major player to reach the solution.20 We effectively anticipatedlengthy negotiations. We correctly forecasted that the West would com-promise on demands for more extreme Kosovar autonomy and with-drawing NATO's initial demands for full autonomy. However, we cannever be certain about some details. For example, as suggested above, theanti-war coalition in the US Congress may or may not have played a centralrole in the settlement, and Milosevic may or may not have granted moreautonomy than the West demanded.21

CONCLUSIONS

Our objective was to determine whether a formal model could allow us toanticipate political developments as the crisis and war in Kosovo unfolded.To see if such an evaluation can be useful, we performed a real-timeanalysis of the Kosovo crisis.22 Our objective was to forecast the outcomeand identify negotiation strategies that could have achieved an effectivesettlement. Let us review our results and contrast them with real outcomes.

First, let us evaluate the positives. The outcome of the Rambouilletnegotiations was anticipated accurately. We forecasted that the last ditchefforts in Paris would fail, and that an outbreak of hostilities was likely tofollow. The evaluations of the outcome of the war were also generallyaccurate. Our analysis indicated that extending the air war alone wasinsufficient to resolve the conflict. We anticipated that negotiations led byRussia were necessary to end the war. We further anticipated that theRussian option would not be accepted quickly but once agreement wasreached, a stable solution was likely. To achieve this solution, the Westwould have to limit demands for full autonomy—and it did.

The analysis also identified options that we cannot verify. The modelindicated that the Russians could have been major dealmakers at Ram-bouillet. We can never know if this was correct The model also deter-mined that the US Congress would play a major role in the air warsettlement. We have no evidence of such activity. The model anticipated

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14

168 B. EFIRD ETAL.

some disagreement among neighbors regarding the Kosovo solution.Again, we see little indication of such opposition.

In general, the expected utility decision model quite accurately anticip-ated the events that led to and followed the seulement of the crisis inKosovo. Could such an analysis have added to policy making during thecrisis? We believe that the answer is a resounding yes. Would suchknowledge have affected outcomes? We will let the reader judge.

This analysis allows us to contrast anticipated outcomes resulting fromnegotiations and war. Figure 6 compares the median outcome supportedby the veto players on the autonomy issue that could have been achievedat Rambouillet with the actual and forecasted solution achieved after theair war.

Figure 6 shows that at Rambouillet, negotiators could have agreed tosubstantial autonomy for Kosovo, while after the air war only limitedautonomy was granted to Kosovo. The West, rather than the Serbians,made concessions. Indeed, NATO disbanded the KLA forces and nowopposes attempts to develop an entity independent of Yugoslavia, while atRambouillet Western powers supported elections within five years todetermine Kosovo's future. The referendum for Kosovar autonomy was acasualty of war.

Recall that data for deployment were not collected during the air warbecause NATO demanded military presence. Recall further that the Ser-bians were unwilling to accept any armed forces on their territory at

Fill Autonomy

Ural«! Autonomy

Wartime Outcome

Figure 6 Potential Kosovar Autonomy: Rambouillet versus Wartime Simulations

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14


NATOTroop

MixedTroops

OSCEMonitors

No ForeignPresence

Simulated Rambouillet Outcome

Figure 7 Potential Troop Deployment: Rambouillet versus Wartime Simulations

Rambouillet. A direct comparison is therefore more difficult However,had a Russian solution emerged in the French meetings, we find thatNATO still would have been directly involved in securing stability forKosovo. The difference is that with the unrealized Rambouillet resolution,Serbian forces could not have been excluded (see Figure 3). What the airwar achieved, compared to our potential Rambouillet agreement, is theexpulsion of Serbian troops from Kosovo.

In sum, the expected utility approach identified a negotiated solutionthat, from the perspective of NATO, is inferior on the issue of autonomybut superior on troop deployment Can one then argue that the air war wasworth fighting? Had negotiators known of the eventual tradeoff, wouldpolicy practitioners have allowed Russian negotiators to lead at Ram-bouillet? Consider the implications of these alternatives.

Systematic analysis can assess the consequences of different answers tothese questions. Had negotiations in France succeeded, then the forcedmigration of ethnic Albanians from Kosovo, as well as the loss of humanUfe and resources associated with the air war, would have been averted orminimized.23 The negotiated agreement would have preserved Yugoslaviaas a single entity and forced the participants to find political accom-modation within Kosovo, or forced a partition after five years of trying.Contrast this outcome to the observed consequences of the successful airwar. In addition to the human cost of the Albanian exodus and the costs of

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14

170 B. EFIRD ETAL.

the aerial campaign, we are now witnessing intensified retaliation byethnic Albanians against Serbians. The ethnic rivalry has not diminished, ithas escalated. Moreover, Kosovo has been defacto partitioned by. 50,000foreign troops, while any commitment to autonomy without a foreignpresence has disappeared.

Our analysis suggests that an outcome similar to the final resolution ofthe Kosovo crisis could have been achieved in negotiations prior to thebombing campaign. The model allows us to simulate alternate strategies,so that the outcome of each strategy may be explored and the mostdesirable resolution to a crisis may be achieved. Despite the presence ofNATO forces, the structural rifts within Yugoslavia have widened. Nofeasible domestic solution to the underlying conflict has emerged. On thepositive side, the international community has imposed stability. We letthe reader decide whether that is sufficient.

Expected utility theory provides the means to forecast and even improvenegotiation and policy strategies. Negotiations, however, seldom changethe underlying structure of decision making constraints. The insights pro-vided by the expected utility model help us to resolve a particular issue inconflict, but do not assure that the structural underpinnings of rivalry areremoved. Systematic analysis can aid policy makers, but cannot stem thetide of structurally caused events. It cannot determine the goals of policy; itcan merely identify better ways to achieve goals. The promulgation of suchanalytical tools within the policy community would refocus the foreignpolicy debate, channeling efforts from policy making under constraints todefining the policy goals that best serve national interests.

APPENDIX: DATA

TABLE laPreference Scale for Autonomy of

Kosovo: Rambouillet Talks

Position Definition

0 No self-rule50 Limited autonomy

100 Autonomy120 Full independence for Kosovo

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14


TABLE IbData Autonomy of Kosovo: Rambouillet Talks

Stakeholder

LDK—Rugova (Kosovo)KLA—Kosovo Liberation Army (Kosovo)Orthodox ChurchAlbanian Diaspora (Kosovo)Milosevic (Serbia)Serbians in Kosovo (Serbia)Yugoslav Army'(Serbia)Seselj—Radical party (Serbia)Other extremist groups (Serbia)Democratic Opposition/Indep media (Serbia)Nano (Albania)Berisha (Albania)Djukanovic (Motenegro)Bulatovic (Montenegro)US GovernmentUS CongressFranceUnited KingdomGermanyItalyUN Secretary GeneralUN Security CouncilNATORussiaMacedonia

Resources

1530105

10015151055

255

255

2001007560301010201050

2

Position

10010040

100400

4000

4010010050408090758080808075805050

Salience

10010010010090

10080

10010050255065607525757065653525806065

TABLE HaPreference Scale for Troop Deployment: Rambouillet Talks

Position Definition

0 No foreign presence30 OSCE monitors31-99 Mixture of troops, increasing to more foreign troops100 NATO troops

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14

172 B. EFIRD ETAL.

TABLE m»Troop Deployment: Rambouillet Talks

Stakeholder

LDK—Rugova (Kosovo)

Resources

15KLA—Kosovo liberation Army (Kosovo) 30Orthodox ChurchAlbanian Diaspora (Kosovo)Milosevic (Serbia)Serbians in Kosovo (Serbia)Yugoslav Army (Serbia)Seselj—Radical party (Serbia)Other extremist groups (Serbia

105

100151510

) 5Democratic Opposition/Indep media (Serbia) SNano (Albania)Berisha (Albania)Djukanovic (Motenegro)Bulatovic (Montenegro)US GovernmentUS CongressFranceUnited KingdomGermanyItalyUN Secretary GeneralUN Security CouncilNATORussiaMacedonia

•

255

255

20010075603010102010502

TABLE m aPreference Scale for Autonomy of

Kosovo: Daring the Air War

Position

03050

100120

Definition

No self-ruleVery limited autonomyLimited autonomyAutonomy

Position

100100

0100

000000

100100500

10040

10010010010010080

10050

100

Full independence for Kosovo

Salience

100100605050607080801060602540

10090

1001001001006060

1006075

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14


TABLE m bAutonomy of Kosovo Daring the Air War

Stakeholder

Berisha (Albania)Albanian GovernmentBulatovic (Montenegro)Djukanovic (Montenegro)US Congress—Pro-War FactionUS Congress—Anti-War FactionUS Executive/GovernmentFranceGermanyItalyMacedoniaNATORussiaUnited KingdomUN Security CouncilUN Secretary GeneralDemocratic OppositionSerbian ExtremistsMilosevicSerbians in KosovoSeselj (Radical Party)Yugoslav ArmyOrthodox ChurchAlbanian DiasporaKosovo Liberation ArmyRugova (LDK)GreeceBulgariaRomaniaHungaryCroatiaSerbians in BosniaFederated Bosnia

Resources

120

1502025

300757575

100100100805

101

40200

52050105

301

40151015101010

Position

100100

055

1000

10010010010010010042

1005050

0000000

10010010080

10010010010035

100

Salience

7010070

100505090909090

10010080909090

10010010010010010010010010010090958585859080

NOTES

1. The expected utility approach has been documented to be over 90% accurate in its forecastsby several sources, including Bueno de Mesquita and Stokman (1994) and Feder (199S).

2. This portion of the analyses was presented at the National War College in Washington,D.C. on February 17, 1999.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14

174 B. EFIRD ETAL

3. We appreciate Decision Insights, Inc.'s willingness to allow the use of their model in thisanalysis. The actual decision analytic tool used for the forecasts utilizes both game the-oretic and decision theoretic calculations. Details on the calculations utilized for suchanalysis are contained in Bueno de Mesquita (1997).

4. "Votes" in this case refer to the influence that may be exerted by stakeholders.5. Ambassador Galbraith kindly consented to provide the data. He read the article and con-

sidered the outcomes valuable. He does not, however, agree with all the assessments made bythe model and is not responsible for the negotiation proposals outlined on the basis of his data.

6. Veto players are simply those stakeholders that can unilaterally prevent the successfulimplementation of an agreement

7. International attention was drawn to Kosovo in 1981 as violence erupted among ethnicAlbanians and the Serbian-controlled police and military forces (Pipa and Repishti, 1984).The conflict simmered, increasing in intensity by 1987, when Milosevic gained the upperhand in Yugoslavia and retracted the autonomy previously granted to Kosovo because of hisdesire to create a Greater Serbian nation. Ethnic Albanians, an ethnic minority in the provinceof the larger Serbia, comprise the majority in Kosovo (Bennett, 1995). Viewing Kosovo as anintegral part of Yugoslavia, Milosevic sought to regain full control by expelling, attacking,and killing the resident Albanians and substituting them with a Serbian population. TheDemocratic League of Kosovo (LDK) was formed in December 1989 and declared its desirefor autonomy and long-term goal of independence for Kosovo. Ibrahim Rugo va was electedas the LDK'sleader in September 1991. Rugova took a softline approach to fulfilling the goalsof the LDK. He did not secure full autonomy for Kosovo, but guided the organization towardcontinued peaceful calls for independence. Since the LDK's method of protest was non-violent, Milosevic tolerated them. This changed with the formation of the Kosovo LiberationArmy (KLA) in November 1997, an organization that is overtly violent in its approach to theindependence movement The KLA put increasing pressure on the LDK to demonstrateAlbanian unity, and forced it to support a more violent position, which drew in Milosevic andescalated the situation to the crisis of February 1999. In this context, the international com-munity began to pay attention to the increasing amount of ethnically-defined violence inKosovo. Tensions mounted in 1998 when Kosovar rebels clashed with Serbian police andmilitary forces. Despite United Nations condemnations and a diplomatic observer force fromthe European Union, over 300,000 Kosovars were displaced from their homes. In September,the UN Security Council demanded that hostilities cease, backed by a threat of NATO airstrikes for non-compliance. This ultimatum produced an agreement that allowed unarmedmonitors from the Organization for Security and Cooperation in Europe (OSCE) to enterKosovo. The peace failed to last, and in January 1999 another45 ethnic Albanians were killedby Serbian forces in the village of Racak, after which OSCE monitors were denied access tocritical areas in Kosovo. The situation intensified as the KLA began launching an offensiveagainst Serbian forces. Thereafter, attacks against the ethnic Albanians increased in fre-quency and intensity, setting the stage for the time period we analyze in this paper.

8. This prediction and other conclusions were contained in a briefing paper presented at theNational War College. These conclusions were contested as many observers felt thatnegotiations could succeed.

9. The model calculates the expected outcome of interactions from the perspective of eachstakeholder in every pairwise combination. Since the model takes such a perceptualapproach to negotiations, we can identify those instances when one stakeholder may notrealize that they have the ability to persuade another stakeholder to change their position(i.e. when the stakeholder does not realize that the expected payoffs of its opponent arehigher for concession than in holding a firm negotiating posture).

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14


10. Recall, however, that all winset solutions are prone to counteroffers within the solutionspace, therefore, even if identified this outcome is not necessarily stable—counterofferswithin the solution space can be credibly made. The winset in gray is the area of overlapbetween all stakeholders' indifference curves and represents the set of all offers that wouldbe accepted by all parties. Even though any point in this area is equivalent, we anticipate thatall parties would maximize their own policy, and select points closes to their own positions.

11. It is not surprising that the US and France would be singled out as deal-breakers inretrospect At the time we performed our analysis, we did not know that all NATOmilitary decisions were determined only by the US, UK, and France. However, after theend of the war, it was revealed that the leadership in these three nations made alldecisions in the bombing campaign (Priest, 1999b). The model's findings suggest thatMilosevic and other Serbian actors realized the source of military decision making andwere unwilling to compromise with potential wartime enemies.

12. This two-dimensional analysis was also conducted for the broader set of stakeholders.Results are similar but the drawing is more complex. A solution emerges as long as theUnited States government and France are not directly involved in the negotiations.

13. Prior to the initiation of conflict, the population of Kosovo, estimated at about twomillion, was 80 percent ethnic Albanian. By late April 1999, one million Kosovars hadbeen displaced or had become refugees in the surrounding region. An additional 40,000have taken asylum in nations of the European Union (EU) and the United States. Theestimated number of missing or dead during the war ranged from 10,000 to 100,000(Economist, 1999). Recent assessments suggest that the lower number is correct Fol-lowing the war, Serbs have started to leave Kosovo despite NATO's presence. Thisexodus has thus far reduced the Serb population in Kosovo by approximately 50%.

14. Once the air war began. Ambassador Galbraith identified two distinct coalitions withinCongress, so that we split the formerly unified actor into an anti-war coalition and a pro-war coalition.

15. Extensive sensitivity analyses were performed on both position and salience to ensure thestability of our forecasts. We allowed the positions of all critical stakeholders to vary byup to 25% and re-computed our forecasts for each variation. In each of the 125 sensitivitysimulations on position, the forecast, position of Milosevic, and position of the westerncoalition varied by less than 2%. Sensitivity analyses were also performed by random-izing the salience of stakeholders. This process simulates the potential for some exo-genous shock to affect a set of stakeholders and disrupt the forecast In this case, 94% ofthe 100 simulations produced virtually identical results. Thus, the forecast reported isrobust to changes in the data.

16. During the Rambouillet negotiations we examined the implications of the potentialremoval of Milosevic from power, as could, for example, have resulted from a widelyanticipated coup. At that stage, the removal of Milosevic produced a very quick com-promise where the Serbians groups agreed to Kosovars' bid for autonomy. However,during the analysis of the air war, we found that the Serbian position stiffened. Even ifMilosevic were no longer in power, once the air war started Serbian unity did not vary.The war strengthened the political position of the more extreme factions.

17. The model also identified a NATO solution that required a split in the coalition andstrategic deception among the allies themselves. For details see Efird, Galbraith, Kuglerand Abdollahian (1999).

18. It has been suggested recently that the Russians and Milosevic had struck a "back door"deal on the issue of troops. If true, such a deal would have motivated Milosevic to backdown during the air campaign, expecting the Russians to delay US troop deployment

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14

176 B. EFIRD ETAL.

with the "airport crisis." Our analysis suggests, however, that the deal forged betweenRussia and Milosevic would only be credible if the US were willing to shin in favor of itin the third round. Indeed, the Russian pressure, applied to Milosevic independently ofthe US, appears to have been critical to forcing the Serbian compromises that led to theend of the air war. Thus, an initial "back door" deal was needed to resolve the crisis.

19. Our fit is not perfect Greece, Albania's President, and the pro-war coalition in the UnitedStates were expected to oppose this outcome. When reviewing the actual record ofevents, we found no such opposition.

20. Indeed, private negotiations between Finnish President Martti Ahtisaari and Russianenvoy Viktor Chernomyrdin ended the war by securing Serbian concessions on die issueof autonomy. These negotiations proceeded without direct US involvement during the airwar (Priest, 1999a).

21. This analysis was completed before the bombing of the Chinese embassy but simulationsindicated that this event would not alter fundamentally the options described here.

22. Pevehouse and Goldstein (1999) also attempt to build a real-time forecasting tool that canbe used by policy analysts. Their efforts focus on pattern recognition using VAR toassess changes in levels of cooperation and competition, and they find that NATObombing would not affect Serbian actions toward Kosovo. This is a similar finding toours, but in the end their approach can only predict the general response patterns in termsof degree of cooperation and conflict, as opposed to the specific stakeholder shifts anddetailed strategic information provided by our approach.

23. Statistics on the number of deaths caused by the air war are hard to obtain, but Yugoslavofficials indicate that NATO bombing killed approximately 600 military and policepersonnel. International organizations and the media in Yugoslavia estimate that between500 and 1,000 civilians were killed during the course of bombing as well (Priest, 1999b).

REFERENCES

Abdollahian, Mark and Carole Alsharabati (1995). "The Middle East: Israel, the Arabs andthe Prospects for Peace." Presented at the Annual Meeting of the American PoliticalScience Association, Chicago.

Bennett, Christopher (1995). Yugoslavia's Bloody Collapse. New York: New York UniversityPress.

Black, Duncan (1958). The Theory of Committees and Elections. Cambridge: CambridgeUniversity Press.

Bueno de Mesquita, Bruce (1997). "A Decision Making Model." International Interactions,VoL 23, Nos. 3-4, pp. 235-266.

Bueno de Mesquita, Bruce, David Newman and Alvin Rabushka (1996). Red Flag overHong Kong. Chatham: Chatham House Publishers.

Bueno de Mesquita and Frans Stokman (eds.) (1994). European Community Decision Mak-ing: Models, Applications, and Comparisons. New Haven: Yale University Press.

Economist. (1999). "The Great Exodus." April 24, p. 48.Efird, Brian, Peter Galbraith, Jacek Kugler and Mark Abdollahian (1999). "Negotiating Peace

in Kosovo," Paper presented at the Conference on Collective Decision and PolicyMaking, University of Groningen, June 21-22.

Feder, Stanley (1995). "Factions and Policon: New Ways to Analyze Politics." In BradfordWesterfield (ed.), Inside CIA's Private World: Declassified Articles from the Agency'sInternal Journal, 1955-1992. New Haven: Yale University Press.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14


Hinich, J. Melvin and Michael C. Munger (1997). Analytical Politics. Cambridge: CambridgeUniversity Press.

Kugler, Jacek and Yi Feng (eds.) (1997). "Special Issue: The Expected Utility Approach toPolicy Decision Making: Assessments, Forecasts, and Strategies." International Inter-actions, Vol. 23, Nos. 3-4, pp. 233-394.

Pevehouse, C Jon and Joshua S. Goldstein (1999). "Serbian Compliance or Defiance inKosovo?" Journal of Conflict Resolution, Vol. 43, No. 4, pp. 538-46.

Pipa, Arshi and Sami Repishti (eds.) (1984). Studies in Kosovo. Boulder: East EuropeanMonographs.

Priest, Dana (1999a). "Kosovo Land Threat May Have Won War." Washington Post,September 19, p. A1.

Priest, Dana (1999b). "France Played Skeptic on Kosovo Attacks." Washington Post,September 20, p. A1.

Priest, Dana (1999c). "Target Selection was Long Process." Washington Post, September 20,p. A l l .

Priest, Dana (1999d). "United NATO Front Was Divided Within." Washington Post,September 21, p. Al.

Weingast, Barry (1979). "A Rational Choice Perspective on Congressional Norms." Amer-ican Journal of Political Science, Vol. 23, pp. 245-62.

CONTRIBUTORS

Brian Efird is completing his Ph.D. in political science at the ClaremontGraduate University and is an Associate at Decision Insights, Inc. in NewYork. His research focuses on interstate conflict, bargaining problems,dispute mediation, and the political economy of growth. His dissertationdeals with the dynamic relationship between war, peace, and integration.In addition to scholarly journal publications, he is the co-author of PowerTransitions. Contact Address: Decision Insights, Inc., 60 East 42nd Street,Suite 514, New York, NY 10165.

Peter W. Galbraith currently is serving as a senior representative of feeUnited Nations in East Timor. From 1993-1998, he held the position ofAmbassador to Croatia, where he played a principal role in devising andimplementing the strategy that ended the 1993-1994 Muslim-Croat war,and where he helped negotiate the Washington Agreement that establishedthe Federation of Bosnia-Herzegovina. Ambassador Galbraith also hasserved as senior advisor to the Senate Foreign Relations Committee. Heholds an A.B. from Harvard College, an M.A. from Oxford University, anda J.D. from Georgetown University Law Center.

Jacek Kugler is the Elisabeth Helms Rosecrans Professor of InternationalRelations at the School of Politics and Economics, Claremont Graduate

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14

178 B. EFIRD ETAL.

University, where he has also served as Director and Chairman. He isthe co-founder of Decision Insights, Inc. His publications in world politicsand political economy are widely available in scholarly journals. He is theco-author of The War Ledger, Births, Deaths, and Taxes; and PowerTransitions; as well as the co-editor of Parity and War; The Long TermStability of Deterrence; and Political Capacity and Economic Behavior.Contact Address: Claremont Graduate University, School for Politics andEconomics, 170 East 10th Street, Claremont, CA 91711-6163.

Mark Andrew Abdollahian is Vice President of Decision Insights, Inc.,a New York consulting firm that forecasts political and economic events.He has lectured extensively on decision making and foreign policy. Hisresearch focuses on leadership, strategy, tactics, the dynamics of interstaterelations, and the historic evaluation of international politics. In addition toscholarly publications on the politics of Russia, Korea and South Africa,he is the co-author of Power Transitions. Contact Address: DecisionInsights Inc., 60 East 42nd Street, Suite 514, New York, NY 10165.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

8:29

17

Dec

embe

r 20

14

negotiating peace in kosovo

Documents